During the formation of a semiconductor device such as a memory device, logic device, microprocessor, etc., several optical lithography steps are typically required. Each optical lithography step typically includes the formation of a blanket photoresist (resist) layer, exposing portions of the resist layer to light using a mask or reticle, removing the exposed resist portions (or the unexposed resist portions if negative resist is used), etching the underlying layer using the resist as a pattern, then stripping the resist. To remove the resist, a high-temperature ash step is performed, then the wafer surface is exposed one or more times to an acid, typically a mixture of hydrogen peroxide (H2O2) and sulfuric acid (H2SO4), often referred to as a “piranha” process, to remove the resist ash which comprises organic resins and metallic contaminants.
Optical lithography adds significantly to the cost of semiconductor device production. Each optical lithography step requires significant time, as the wafers must be moved from a station that deposits the resist, typically a spin-coat process, to a stepper which exposes the resist using a mask or reticle. After the exposed or unexposed resist is removed, the wafer is moved to an etcher to etch the underlying layer, then to a furnace that ashes the resist, and finally to a piranha bath to remove the ashed resist. Optical lithography also adds expense to the wafer as it requires materials including resist and acids and their eventual disposal, and also may decrease yields from misalignment of the mask.
A continual design goal during the manufacture of semiconductor devices is to produce smaller features. One limit to this goal is the deficiencies in optical lithography that restrict the minimum feature size. This minimum for feature sizes results from various optical properties of the optical lithographic process.
To overcome the deficiencies of optical lithography, research is ongoing into other patterning techniques. One such technique is nanoimprinting, which may be classified into the three categories listed below.
A first nanoimprinting technique, “hot embossing” or “thermal embossing,” comprises the use of a substrate to be patterned and a liquid coating, typically a low-viscosity monomer, formed over the substrate. A template, which comprises a surface with a raised pattern on the surface, is pressed into the coating, then the coating is cured by heating. The template is removed and the coating is used as a mask to etch the substrate.
In a second nanoimprinting technique, “UV nanoimprinting,” a transparent template is pressed into a UV-curable coating over the surface of the substrate to be patterned, then the coating is exposed to UV wavelength light flashed through the transparent template to cure the coating. The template is removed and the substrate is patterned using the coating as a mask. It is generally believed that UV nanoimprinting is the most likely candidate for semiconductor processing.
With a third nanoimprinting technique, “micro contact printing,” the coating is applied to the pattern on a soft, flexible template, then the template is pressed onto the substrate. The coating adheres to the substrate, then the template is removed and the coating is used as a mask to etch the substrate. Because the template is flexible, it is difficult to print features that are as small as those printed with the other two techniques.
The template used for nanopatterning may be formed using any of several methods. For example, molecular beam epitaxy (MBE) may be used to create a physical template for nanowire patterning. This method enables simple physical transfer of fully formed metallic wires from a selectively etched superlattice, for example GaAs/Al0.8Ga0.2As, onto a silicon wafer. The nanowires are defined by evaporating metal directly onto the GaAs layers of the superlattice after selective removal of the AlGaAs to create voids between the GaAs layers. By depositing the metal solely on the GaAs, the wire widths are defined by the thickness of the GaAs layers, and the separation between the wires is defined by the thickness of the AlGaAs layers. Atomic-level control over the thickness and composition of each layer is achieved by synthesizing the GaAs/AlGaAs superlattice via MBE. In this manner, automatically defined templates for metal wires can be fabricated with widths of 1 nanometer (nm) or less, although wires of this dimension have not been successfully transferred thus far.
Another method for forming the nanopatterning template uses photocurable nanoimprint lithography (P-NIL). A mold is pressed into a low viscosity photocurable resist liquid to physically deform the resist shape such that it conforms to the topology of the mold. The various components in the liquid resist are crosslinked through exposure to UV light to produce a uniform, relatively rigid polymer network. The mold is then separated from the cured resist, then an anisotropic reactive ion etch (RIE) is performed to remove the residual resist in the compressed area, thereby exposing the substrate surface.
Other methods exist for creating nanowires of small dimensions (less than 20 nm), but organizing these wires into highly ordered arrays with predetermined spacing and registry has been challenging. For practical technological applications, it is necessary not only to have nanoscale wire widths but also to know the precise location and registry between wires. Otherwise, to make contact to the nanowires, the beginning and ending locations of every nanowire must be determined and the contact patterned specifically for each wire.
Various problems may be encountered during conventional template formation. For example, resolution is limited by proximity effects inherent with electron beam lithography and thus patterning sub-35 nm pitch features is difficult. The MBE technique discussed above can create a small cluster of regularly spaced wires, but reliably creating such wires over the entire surface of the substrate is not possible using present MBE techniques.
A method for forming a template for nanoimprinting which may overcome various problems previously encountered during template manufacture, and various methods of use for the template, would be desirable.